Stealth SUSY Analysis Instructions

Carnegie Mellon Group

(M. Andrews, N. Bower, T. Mudholkar, M. Paulini, M. Sun, M. Weinberg)

For full details about the compulsory and optional arguments to pass to any executable (including the C++ executables!), please do: executableName -h; for example, ./mergeFiles.py -h

Creating N-tuples

Running ntuplizer

Please see: ggNtuplizer with modifications

The crab utility is documented here.

Event selection

To run selection: ./submitEventSelectionJobs.py --optionalIdentifier example_identifier

By default this won't submit the jobs, only create the jdl files which you can examine. To submit the jobs, pass the flag --isProductionRun.

Outputs will be stored in the folder /store/user/lpcsusystealth/selections/DoublePhoton<_optional_identifier_if_set>

Merging output files

To merge all the files created by the selection script: ./runSelectionMerge.py --optionalIdentifier example_identifier

Full analysis chain

To run full analysis chain: ./runAnalysis.py --selectionSuffix <example_selection_identifier> --optionalIdentifier <example_analysis_identifier>

To run a specific chain, do: ./runAnalysis.sh --chain "type".

If the flag --runUnblinded is passed, the chain runs with the signal region unblinded: the observed data for the signal sample is now plotted as well, and observed limit contours are drawn in addition to the expected limit contours.

• Chain data: This chain examines ST distributions in the control and signal region data. This outputs the uncertainty estimates including the ST scaling estimate from the control data, the expected NEvents distributions, and also the number of events recorded (outside the blinded region).
• Chain MC: This chain examines ST distributions in the signal region MC samples. This chain runs over the full signal MC n-tuples and generates histograms of the expected number of events in each (m_gluino, m_neutralino) bin, as well as the expected number of events with the various shifts. In addition, this chain generates MC systematics (luminosity, statistical uncertainty on number of passing events, JEC, unclustered MET, JER, prefiring, photon scale factor) using the distributions created in the data chain.
• Chain combine: Using the expected and observed background number of events from the data chain, the expected number of signal events from the MC chain, and all systematic uncertainties from both chains, this chain creates three data cards per (m_gluino, m_neutralino) bin: with the nominal cross section and with it shifted up and down by the theoretical uncertainty. In addition, this chain submits the Condor jobs to run the combine tool on these datacards.
• Chain signalContamination: This chain generates histograms of the potential signal contamination throughout the (m_gluino, m_neutralino) phase-space.
• Chain ancillaryPlots: This chain generates publication-quality histograms of the expected number of events (for the signal sample) and the observed number of events (for the control sample) with a few useful distributions from MC overlaid.
• Chain limits: This chain generates the 95% expected limit plots with contours at signal_strength = 1.